Because requirements for establishing spatial data for location-based services (LBS) have increased in demand together with an increase in the number of complex structures, especially in urban areas, research has revisited the limitations of data models in representing space. Though research and corresponding applications continue to explore indoor spaces, their complete and accurate representation remains a challenge. Indoor space presents a hierarchical structure, but, unlike their topological relationships, data models have overlooked this. As subspacing presents a method to express this hierarchy of space, we aimed to develop a subspacing framework for expressing topological and hierarchical relationships at various levels of indoor space. We accomplished this by investigating the hierarchy of indoor space structures and how this relates to implementing a multi-level Node-Relation Structure (NRS) representation of indoor space through subspacing. Furthermore, we formalized these concepts by extending the IndoorGML core model. Then, we demonstrated the potential of the proposed framework through an experiment on sample data by generating corresponding network representations at different levels of detail.